1 | /**************************************************************************** |
2 | ** |
3 | ** Copyright (C) 2018 Intel Corporation |
4 | ** |
5 | ** Permission is hereby granted, free of charge, to any person obtaining a copy |
6 | ** of this software and associated documentation files (the "Software"), to deal |
7 | ** in the Software without restriction, including without limitation the rights |
8 | ** to use, copy, modify, merge, publish, distribute, sublicense, and/or sell |
9 | ** copies of the Software, and to permit persons to whom the Software is |
10 | ** furnished to do so, subject to the following conditions: |
11 | ** |
12 | ** The above copyright notice and this permission notice shall be included in |
13 | ** all copies or substantial portions of the Software. |
14 | ** |
15 | ** THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
16 | ** IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
17 | ** FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE |
18 | ** AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
19 | ** LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, |
20 | ** OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN |
21 | ** THE SOFTWARE. |
22 | ** |
23 | ****************************************************************************/ |
24 | |
25 | #ifndef _BSD_SOURCE |
26 | #define _BSD_SOURCE 1 |
27 | #endif |
28 | #ifndef _DEFAULT_SOURCE |
29 | #define _DEFAULT_SOURCE 1 |
30 | #endif |
31 | #ifndef __STDC_LIMIT_MACROS |
32 | # define __STDC_LIMIT_MACROS 1 |
33 | #endif |
34 | |
35 | #include "cbor.h" |
36 | #include "cborinternal_p.h" |
37 | #include "compilersupport_p.h" |
38 | |
39 | #include <string.h> |
40 | |
41 | /** |
42 | * \defgroup CborParsing Parsing CBOR streams |
43 | * \brief Group of functions used to parse CBOR streams. |
44 | * |
45 | * TinyCBOR provides functions for pull-based stream parsing of a CBOR-encoded |
46 | * payload. The main data type for the parsing is a CborValue, which behaves |
47 | * like an iterator and can be used to extract the encoded data. It is first |
48 | * initialized with a call to cbor_parser_init() and is usually used to extract |
49 | * exactly one item, most often an array or map. |
50 | * |
51 | * Nested CborValue objects can be parsed using cbor_value_enter_container(). |
52 | * Each call to cbor_value_enter_container() must be matched by a call to |
53 | * cbor_value_leave_container(), with the exact same parameters. |
54 | * |
55 | * The example below initializes a CborParser object, begins the parsing with a |
56 | * CborValue and decodes a single integer: |
57 | * |
58 | * \code |
59 | * int extract_int(const uint8_t *buffer, size_t len) |
60 | * { |
61 | * CborParser parser; |
62 | * CborValue value; |
63 | * int result; |
64 | * cbor_parser_init(buffer, len, 0, &parser, &value); |
65 | * cbor_value_get_int(&value, &result); |
66 | * return result; |
67 | * } |
68 | * \endcode |
69 | * |
70 | * The code above does no error checking, which means it assumes the data comes |
71 | * from a source trusted to send one properly-encoded integer. The following |
72 | * example does the exact same operation, but includes error checking and |
73 | * returns 0 on parsing failure: |
74 | * |
75 | * \code |
76 | * int extract_int(const uint8_t *buffer, size_t len) |
77 | * { |
78 | * CborParser parser; |
79 | * CborValue value; |
80 | * int result; |
81 | * if (cbor_parser_init(buffer, len, 0, &parser, &value) != CborNoError) |
82 | * return 0; |
83 | * if (!cbor_value_is_integer(&value) || |
84 | * cbor_value_get_int(&value, &result) != CborNoError) |
85 | * return 0; |
86 | * return result; |
87 | * } |
88 | * \endcode |
89 | * |
90 | * Note, in the example above, that one can't distinguish a parsing failure |
91 | * from an encoded value of zero. Reporting a parsing error is left as an |
92 | * exercise to the reader. |
93 | * |
94 | * The code above does not execute a range-check either: it is possible that |
95 | * the value decoded from the CBOR stream encodes a number larger than what can |
96 | * be represented in a variable of type \c{int}. If detecting that case is |
97 | * important, the code should call cbor_value_get_int_checked() instead. |
98 | * |
99 | * <h3 class="groupheader">Memory and parsing constraints</h3> |
100 | * |
101 | * TinyCBOR is designed to run with little memory and with minimal overhead. |
102 | * Except where otherwise noted, the parser functions always run on constant |
103 | * time (O(1)), do not recurse and never allocate memory (thus, stack usage is |
104 | * bounded and is O(1)). |
105 | * |
106 | * <h3 class="groupheader">Error handling and preconditions</h3> |
107 | * |
108 | * All functions operating on a CborValue return a CborError condition, with |
109 | * CborNoError standing for the normal situation in which no parsing error |
110 | * occurred. All functions may return parsing errors in case the stream cannot |
111 | * be decoded properly, be it due to corrupted data or due to reaching the end |
112 | * of the input buffer. |
113 | * |
114 | * Error conditions must not be ignored. All decoder functions have undefined |
115 | * behavior if called after an error has been reported, and may crash. |
116 | * |
117 | * Some functions are also documented to have preconditions, like |
118 | * cbor_value_get_int() requiring that the input be an integral value. |
119 | * Violation of preconditions also results in undefined behavior and the |
120 | * program may crash. |
121 | */ |
122 | |
123 | /** |
124 | * \addtogroup CborParsing |
125 | * @{ |
126 | */ |
127 | |
128 | /** |
129 | * \struct CborValue |
130 | * |
131 | * This type contains one value parsed from the CBOR stream. Each CborValue |
132 | * behaves as an iterator in a StAX-style parser. |
133 | * |
134 | * \if privatedocs |
135 | * Implementation details: the CborValue contains these fields: |
136 | * \list |
137 | * \li ptr: pointer to the actual data |
138 | * \li flags: flags from the decoder |
139 | * \li extra: partially decoded integer value (0, 1 or 2 bytes) |
140 | * \li remaining: remaining items in this collection after this item or UINT32_MAX if length is unknown |
141 | * \endlist |
142 | * \endif |
143 | */ |
144 | |
145 | static uint64_t extract_number_and_advance(CborValue *it) |
146 | { |
147 | /* This function is only called after we've verified that the number |
148 | * here is valid, so we can just use _cbor_value_extract_int64_helper. */ |
149 | uint8_t descriptor; |
150 | uint64_t v = _cbor_value_extract_int64_helper(value: it); |
151 | |
152 | read_bytes_unchecked(it, dst: &descriptor, offset: 0, n: 1); |
153 | descriptor &= SmallValueMask; |
154 | |
155 | size_t bytesNeeded = descriptor < Value8Bit ? 0 : (1 << (descriptor - Value8Bit)); |
156 | advance_bytes(it, n: bytesNeeded + 1); |
157 | |
158 | return v; |
159 | } |
160 | |
161 | static bool is_fixed_type(uint8_t type) |
162 | { |
163 | return type != CborTextStringType && type != CborByteStringType && type != CborArrayType && |
164 | type != CborMapType; |
165 | } |
166 | |
167 | static CborError preparse_value(CborValue *it) |
168 | { |
169 | enum { |
170 | /* flags to keep */ |
171 | FlagsToKeep = CborIteratorFlag_ContainerIsMap | CborIteratorFlag_NextIsMapKey |
172 | }; |
173 | uint8_t descriptor; |
174 | |
175 | /* are we at the end? */ |
176 | it->type = CborInvalidType; |
177 | it->flags &= FlagsToKeep; |
178 | if (!read_bytes(it, dst: &descriptor, offset: 0, n: 1)) |
179 | return CborErrorUnexpectedEOF; |
180 | |
181 | uint8_t type = descriptor & MajorTypeMask; |
182 | it->type = type; |
183 | it->extra = (descriptor &= SmallValueMask); |
184 | |
185 | if (descriptor > Value64Bit) { |
186 | if (unlikely(descriptor != IndefiniteLength)) |
187 | return type == CborSimpleType ? CborErrorUnknownType : CborErrorIllegalNumber; |
188 | if (likely(!is_fixed_type(type))) { |
189 | /* special case */ |
190 | it->flags |= CborIteratorFlag_UnknownLength; |
191 | it->type = type; |
192 | return CborNoError; |
193 | } |
194 | return type == CborSimpleType ? CborErrorUnexpectedBreak : CborErrorIllegalNumber; |
195 | } |
196 | |
197 | size_t bytesNeeded = descriptor < Value8Bit ? 0 : (1 << (descriptor - Value8Bit)); |
198 | |
199 | if (bytesNeeded) { |
200 | if (!can_read_bytes(it, n: bytesNeeded + 1)) |
201 | return CborErrorUnexpectedEOF; |
202 | |
203 | it->extra = 0; |
204 | |
205 | /* read up to 16 bits into it->extra */ |
206 | if (bytesNeeded == 1) { |
207 | uint8_t ; |
208 | read_bytes_unchecked(it, dst: &extra, offset: 1, n: bytesNeeded); |
209 | it->extra = extra; |
210 | } else if (bytesNeeded == 2) { |
211 | read_bytes_unchecked(it, dst: &it->extra, offset: 1, n: bytesNeeded); |
212 | it->extra = cbor_ntohs(it->extra); |
213 | } else { |
214 | cbor_static_assert(CborIteratorFlag_IntegerValueTooLarge == (Value32Bit & 3)); |
215 | cbor_static_assert((CborIteratorFlag_IntegerValueIs64Bit | |
216 | CborIteratorFlag_IntegerValueTooLarge) == (Value64Bit & 3)); |
217 | it->flags |= (descriptor & 3); |
218 | } |
219 | } |
220 | |
221 | uint8_t majortype = type >> MajorTypeShift; |
222 | if (majortype == NegativeIntegerType) { |
223 | it->flags |= CborIteratorFlag_NegativeInteger; |
224 | it->type = CborIntegerType; |
225 | } else if (majortype == SimpleTypesType) { |
226 | switch (descriptor) { |
227 | case FalseValue: |
228 | it->extra = false; |
229 | it->type = CborBooleanType; |
230 | break; |
231 | |
232 | case SinglePrecisionFloat: |
233 | case DoublePrecisionFloat: |
234 | it->flags |= CborIteratorFlag_IntegerValueTooLarge; |
235 | /* fall through */ |
236 | case TrueValue: |
237 | case NullValue: |
238 | case UndefinedValue: |
239 | case HalfPrecisionFloat: |
240 | read_bytes_unchecked(it, dst: &it->type, offset: 0, n: 1); |
241 | break; |
242 | |
243 | case SimpleTypeInNextByte: |
244 | #ifndef CBOR_PARSER_NO_STRICT_CHECKS |
245 | if (unlikely(it->extra < 32)) { |
246 | it->type = CborInvalidType; |
247 | return CborErrorIllegalSimpleType; |
248 | } |
249 | #endif |
250 | break; |
251 | |
252 | case 28: |
253 | case 29: |
254 | case 30: |
255 | case Break: |
256 | cbor_assert(false); /* these conditions can't be reached */ |
257 | return CborErrorUnexpectedBreak; |
258 | } |
259 | } |
260 | |
261 | return CborNoError; |
262 | } |
263 | |
264 | static CborError preparse_next_value_nodecrement(CborValue *it) |
265 | { |
266 | uint8_t byte; |
267 | if (it->remaining == UINT32_MAX && read_bytes(it, dst: &byte, offset: 0, n: 1) && byte == (uint8_t)BreakByte) { |
268 | /* end of map or array */ |
269 | if ((it->flags & CborIteratorFlag_ContainerIsMap && it->flags & CborIteratorFlag_NextIsMapKey) |
270 | || it->type == CborTagType) { |
271 | /* but we weren't expecting it! */ |
272 | return CborErrorUnexpectedBreak; |
273 | } |
274 | it->type = CborInvalidType; |
275 | it->remaining = 0; |
276 | it->flags |= CborIteratorFlag_UnknownLength; /* leave_container must consume the Break */ |
277 | return CborNoError; |
278 | } |
279 | |
280 | return preparse_value(it); |
281 | } |
282 | |
283 | static CborError preparse_next_value(CborValue *it) |
284 | { |
285 | /* tags don't count towards item totals or whether we've successfully |
286 | * read a map's key or value */ |
287 | bool itemCounts = it->type != CborTagType; |
288 | |
289 | if (it->remaining != UINT32_MAX) { |
290 | if (itemCounts && --it->remaining == 0) { |
291 | it->type = CborInvalidType; |
292 | it->flags &= ~CborIteratorFlag_UnknownLength; /* no Break to consume */ |
293 | return CborNoError; |
294 | } |
295 | } |
296 | if (itemCounts) { |
297 | /* toggle the flag indicating whether this was a map key */ |
298 | it->flags ^= CborIteratorFlag_NextIsMapKey; |
299 | } |
300 | return preparse_next_value_nodecrement(it); |
301 | } |
302 | |
303 | static CborError advance_internal(CborValue *it) |
304 | { |
305 | uint64_t length = extract_number_and_advance(it); |
306 | |
307 | if (it->type == CborByteStringType || it->type == CborTextStringType) { |
308 | cbor_assert(length == (size_t)length); |
309 | cbor_assert((it->flags & CborIteratorFlag_UnknownLength) == 0); |
310 | advance_bytes(it, n: length); |
311 | } |
312 | |
313 | return preparse_next_value(it); |
314 | } |
315 | |
316 | /** \internal |
317 | * |
318 | * Decodes the CBOR integer value when it is larger than the 16 bits available |
319 | * in value->extra. This function requires that value->flags have the |
320 | * CborIteratorFlag_IntegerValueTooLarge flag set. |
321 | * |
322 | * This function is also used to extract single- and double-precision floating |
323 | * point values (SinglePrecisionFloat == Value32Bit and DoublePrecisionFloat == |
324 | * Value64Bit). |
325 | */ |
326 | uint64_t _cbor_value_decode_int64_internal(const CborValue *value) |
327 | { |
328 | cbor_assert(value->flags & CborIteratorFlag_IntegerValueTooLarge || |
329 | value->type == CborFloatType || value->type == CborDoubleType); |
330 | if (value->flags & CborIteratorFlag_IntegerValueIs64Bit) |
331 | return read_uint64(it: value, offset: 1); |
332 | |
333 | return read_uint32(it: value, offset: 1); |
334 | } |
335 | |
336 | /** |
337 | * Initializes the CBOR parser for parsing \a size bytes beginning at \a |
338 | * buffer. Parsing will use flags set in \a flags. The iterator to the first |
339 | * element is returned in \a it. |
340 | * |
341 | * The \a parser structure needs to remain valid throughout the decoding |
342 | * process. It is not thread-safe to share one CborParser among multiple |
343 | * threads iterating at the same time, but the object can be copied so multiple |
344 | * threads can iterate. |
345 | */ |
346 | CborError cbor_parser_init(const uint8_t *buffer, size_t size, uint32_t flags, CborParser *parser, CborValue *it) |
347 | { |
348 | memset(s: parser, c: 0, n: sizeof(*parser)); |
349 | parser->source.end = buffer + size; |
350 | parser->flags = (enum CborParserGlobalFlags)flags; |
351 | it->parser = parser; |
352 | it->source.ptr = buffer; |
353 | it->remaining = 1; /* there's one type altogether, usually an array or map */ |
354 | it->flags = 0; |
355 | return preparse_value(it); |
356 | } |
357 | |
358 | CborError cbor_parser_init_reader(const struct CborParserOperations *ops, CborParser *parser, CborValue *it, void *token) |
359 | { |
360 | memset(s: parser, c: 0, n: sizeof(*parser)); |
361 | parser->source.ops = ops; |
362 | parser->flags = CborParserFlag_ExternalSource; |
363 | it->parser = parser; |
364 | it->source.token = token; |
365 | it->remaining = 1; |
366 | return preparse_value(it); |
367 | } |
368 | |
369 | /** |
370 | * \fn bool cbor_value_at_end(const CborValue *it) |
371 | * |
372 | * Returns true if \a it has reached the end of the iteration, usually when |
373 | * advancing after the last item in an array or map. |
374 | * |
375 | * In the case of the outermost CborValue object, this function returns true |
376 | * after decoding a single element. A pointer to the first byte of the |
377 | * remaining data (if any) can be obtained with cbor_value_get_next_byte(). |
378 | * |
379 | * \sa cbor_value_advance(), cbor_value_is_valid(), cbor_value_get_next_byte() |
380 | */ |
381 | |
382 | /** |
383 | * \fn const uint8_t *cbor_value_get_next_byte(const CborValue *it) |
384 | * |
385 | * Returns a pointer to the next byte that would be decoded if this CborValue |
386 | * object were advanced. |
387 | * |
388 | * This function is useful if cbor_value_at_end() returns true for the |
389 | * outermost CborValue: the pointer returned is the first byte of the data |
390 | * remaining in the buffer, if any. Code can decide whether to begin decoding a |
391 | * new CBOR data stream from this point, or parse some other data appended to |
392 | * the same buffer. |
393 | * |
394 | * This function may be used even after a parsing error. If that occurred, |
395 | * then this function returns a pointer to where the parsing error occurred. |
396 | * Note that the error recovery is not precise and the pointer may not indicate |
397 | * the exact byte containing bad data. |
398 | * |
399 | * This function makes sense only when using a linear buffer (that is, when the |
400 | * parser is initialize by cbor_parser_init()). If using an external source, |
401 | * this function may return garbage; instead, consult the external source itself |
402 | * to find out more details about the presence of more data. |
403 | * |
404 | * \sa cbor_value_at_end() |
405 | */ |
406 | |
407 | CborError cbor_value_reparse(CborValue *it) |
408 | { |
409 | if (it->flags & CborIteratorFlag_IteratingStringChunks) |
410 | return CborNoError; |
411 | return preparse_next_value_nodecrement(it); |
412 | } |
413 | |
414 | /** |
415 | * \fn bool cbor_value_is_valid(const CborValue *it) |
416 | * |
417 | * Returns true if the iterator \a it contains a valid value. Invalid iterators |
418 | * happen when iteration reaches the end of a container (see \ref |
419 | * cbor_value_at_end()) or when a search function resulted in no matches. |
420 | * |
421 | * \sa cbor_value_advance(), cbor_value_at_end(), cbor_value_get_type() |
422 | */ |
423 | |
424 | /** |
425 | * Performs a basic validation of the CBOR stream pointed by \a it and returns |
426 | * the error it found. If no error was found, it returns CborNoError and the |
427 | * application can iterate over the items with certainty that no other errors |
428 | * will appear during parsing. |
429 | * |
430 | * A basic validation checks for: |
431 | * \list |
432 | * \li absence of undefined additional information bytes; |
433 | * \li well-formedness of all numbers, lengths, and simple values; |
434 | * \li string contents match reported sizes; |
435 | * \li arrays and maps contain the number of elements they are reported to have; |
436 | * \endlist |
437 | * |
438 | * For further checks, see cbor_value_validate(). |
439 | * |
440 | * This function has the same timing and memory requirements as |
441 | * cbor_value_advance(). |
442 | * |
443 | * \sa cbor_value_validate(), cbor_value_advance() |
444 | */ |
445 | CborError cbor_value_validate_basic(const CborValue *it) |
446 | { |
447 | CborValue value = *it; |
448 | return cbor_value_advance(it: &value); |
449 | } |
450 | |
451 | /** |
452 | * Advances the CBOR value \a it by one fixed-size position. Fixed-size types |
453 | * are: integers, tags, simple types (including boolean, null and undefined |
454 | * values) and floating point types. |
455 | * |
456 | * If the type is not of fixed size, this function has undefined behavior. Code |
457 | * must be sure that the current type is one of the fixed-size types before |
458 | * calling this function. This function is provided because it can guarantee |
459 | * that it runs in constant time (O(1)). |
460 | * |
461 | * If the caller is not able to determine whether the type is fixed or not, code |
462 | * can use the cbor_value_advance() function instead. |
463 | * |
464 | * \sa cbor_value_at_end(), cbor_value_advance(), cbor_value_enter_container(), cbor_value_leave_container() |
465 | */ |
466 | CborError cbor_value_advance_fixed(CborValue *it) |
467 | { |
468 | cbor_assert(it->type != CborInvalidType); |
469 | cbor_assert(is_fixed_type(it->type)); |
470 | if (!it->remaining) |
471 | return CborErrorAdvancePastEOF; |
472 | return advance_internal(it); |
473 | } |
474 | |
475 | static CborError advance_recursive(CborValue *it, int nestingLevel) |
476 | { |
477 | CborError err; |
478 | CborValue recursed; |
479 | |
480 | if (is_fixed_type(type: it->type)) |
481 | return advance_internal(it); |
482 | |
483 | if (!cbor_value_is_container(it)) { |
484 | size_t len = SIZE_MAX; |
485 | return _cbor_value_copy_string(value: it, NULL, buflen: &len, next: it); |
486 | } |
487 | |
488 | /* map or array */ |
489 | if (nestingLevel == 0) |
490 | return CborErrorNestingTooDeep; |
491 | |
492 | err = cbor_value_enter_container(it, recursed: &recursed); |
493 | if (err) |
494 | return err; |
495 | while (!cbor_value_at_end(it: &recursed)) { |
496 | err = advance_recursive(it: &recursed, nestingLevel: nestingLevel - 1); |
497 | if (err) |
498 | return err; |
499 | } |
500 | return cbor_value_leave_container(it, recursed: &recursed); |
501 | } |
502 | |
503 | |
504 | /** |
505 | * Advances the CBOR value \a it by one element, skipping over containers. |
506 | * Unlike cbor_value_advance_fixed(), this function can be called on a CBOR |
507 | * value of any type. However, if the type is a container (map or array) or a |
508 | * string with a chunked payload, this function will not run in constant time |
509 | * and will recurse into itself (it will run on O(n) time for the number of |
510 | * elements or chunks and will use O(n) memory for the number of nested |
511 | * containers). |
512 | * |
513 | * The number of recursions can be limited at compile time to avoid stack |
514 | * exhaustion in constrained systems. |
515 | * |
516 | * \sa cbor_value_at_end(), cbor_value_advance_fixed(), cbor_value_enter_container(), cbor_value_leave_container() |
517 | */ |
518 | CborError cbor_value_advance(CborValue *it) |
519 | { |
520 | cbor_assert(it->type != CborInvalidType); |
521 | if (!it->remaining) |
522 | return CborErrorAdvancePastEOF; |
523 | return advance_recursive(it, CBOR_PARSER_MAX_RECURSIONS); |
524 | } |
525 | |
526 | /** |
527 | * \fn bool cbor_value_is_tag(const CborValue *value) |
528 | * |
529 | * Returns true if the iterator \a value is valid and points to a CBOR tag. |
530 | * |
531 | * \sa cbor_value_get_tag(), cbor_value_skip_tag() |
532 | */ |
533 | |
534 | /** |
535 | * \fn CborError cbor_value_get_tag(const CborValue *value, CborTag *result) |
536 | * |
537 | * Retrieves the CBOR tag value that \a value points to and stores it in \a |
538 | * result. If the iterator \a value does not point to a CBOR tag value, the |
539 | * behavior is undefined, so checking with \ref cbor_value_get_type or with |
540 | * \ref cbor_value_is_tag is recommended. |
541 | * |
542 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_tag() |
543 | */ |
544 | |
545 | /** |
546 | * Advances the CBOR value \a it until it no longer points to a tag. If \a it is |
547 | * already not pointing to a tag, then this function returns it unchanged. |
548 | * |
549 | * This function does not run in constant time: it will run on O(n) for n being |
550 | * the number of tags. It does use constant memory (O(1) memory requirements). |
551 | * |
552 | * \sa cbor_value_advance_fixed(), cbor_value_advance() |
553 | */ |
554 | CborError cbor_value_skip_tag(CborValue *it) |
555 | { |
556 | while (cbor_value_is_tag(value: it)) { |
557 | CborError err = cbor_value_advance_fixed(it); |
558 | if (err) |
559 | return err; |
560 | } |
561 | return CborNoError; |
562 | } |
563 | |
564 | /** |
565 | * \fn bool cbor_value_is_container(const CborValue *it) |
566 | * |
567 | * Returns true if the \a it value is a container and requires recursion in |
568 | * order to decode (maps and arrays), false otherwise. |
569 | */ |
570 | |
571 | /** |
572 | * Creates a CborValue iterator pointing to the first element of the container |
573 | * represented by \a it and saves it in \a recursed. The \a it container object |
574 | * needs to be kept and passed again to cbor_value_leave_container() in order |
575 | * to continue iterating past this container. |
576 | * |
577 | * The \a it CborValue iterator must point to a container. |
578 | * |
579 | * \sa cbor_value_is_container(), cbor_value_leave_container(), cbor_value_advance() |
580 | */ |
581 | CborError cbor_value_enter_container(const CborValue *it, CborValue *recursed) |
582 | { |
583 | cbor_static_assert(CborIteratorFlag_ContainerIsMap == (CborMapType & ~CborArrayType)); |
584 | cbor_assert(cbor_value_is_container(it)); |
585 | *recursed = *it; |
586 | |
587 | if (it->flags & CborIteratorFlag_UnknownLength) { |
588 | recursed->remaining = UINT32_MAX; |
589 | advance_bytes(it: recursed, n: 1); |
590 | } else { |
591 | uint64_t len = extract_number_and_advance(it: recursed); |
592 | |
593 | recursed->remaining = (uint32_t)len; |
594 | if (recursed->remaining != len || len == UINT32_MAX) { |
595 | /* back track the pointer to indicate where the error occurred */ |
596 | copy_current_position(dst: recursed, src: it); |
597 | return CborErrorDataTooLarge; |
598 | } |
599 | if (recursed->type == CborMapType) { |
600 | /* maps have keys and values, so we need to multiply by 2 */ |
601 | if (recursed->remaining > UINT32_MAX / 2) { |
602 | /* back track the pointer to indicate where the error occurred */ |
603 | copy_current_position(dst: recursed, src: it); |
604 | return CborErrorDataTooLarge; |
605 | } |
606 | recursed->remaining *= 2; |
607 | } |
608 | if (len == 0) { |
609 | /* the case of the empty container */ |
610 | recursed->type = CborInvalidType; |
611 | return CborNoError; |
612 | } |
613 | } |
614 | recursed->flags = (recursed->type & CborIteratorFlag_ContainerIsMap); |
615 | return preparse_next_value_nodecrement(it: recursed); |
616 | } |
617 | |
618 | /** |
619 | * Updates \a it to point to the next element after the container. The \a |
620 | * recursed object needs to point to the element obtained either by advancing |
621 | * the last element of the container (via cbor_value_advance(), |
622 | * cbor_value_advance_fixed(), a nested cbor_value_leave_container(), or the \c |
623 | * next pointer from cbor_value_copy_string() or cbor_value_dup_string()). |
624 | * |
625 | * The \a it and \a recursed parameters must be the exact same as passed to |
626 | * cbor_value_enter_container(). |
627 | * |
628 | * \sa cbor_value_enter_container(), cbor_value_at_end() |
629 | */ |
630 | CborError cbor_value_leave_container(CborValue *it, const CborValue *recursed) |
631 | { |
632 | cbor_assert(cbor_value_is_container(it)); |
633 | cbor_assert(recursed->type == CborInvalidType); |
634 | |
635 | copy_current_position(dst: it, src: recursed); |
636 | if (recursed->flags & CborIteratorFlag_UnknownLength) |
637 | advance_bytes(it, n: 1); |
638 | return preparse_next_value(it); |
639 | } |
640 | |
641 | |
642 | /** |
643 | * \fn CborType cbor_value_get_type(const CborValue *value) |
644 | * |
645 | * Returns the type of the CBOR value that the iterator \a value points to. If |
646 | * \a value does not point to a valid value, this function returns \ref |
647 | * CborInvalidType. |
648 | * |
649 | * TinyCBOR also provides functions to test directly if a given CborValue object |
650 | * is of a given type, like cbor_value_is_text_string() and cbor_value_is_null(). |
651 | * |
652 | * \sa cbor_value_is_valid() |
653 | */ |
654 | |
655 | /** |
656 | * \fn bool cbor_value_is_null(const CborValue *value) |
657 | * |
658 | * Returns true if the iterator \a value is valid and points to a CBOR null type. |
659 | * |
660 | * \sa cbor_value_is_valid(), cbor_value_is_undefined() |
661 | */ |
662 | |
663 | /** |
664 | * \fn bool cbor_value_is_undefined(const CborValue *value) |
665 | * |
666 | * Returns true if the iterator \a value is valid and points to a CBOR undefined type. |
667 | * |
668 | * \sa cbor_value_is_valid(), cbor_value_is_null() |
669 | */ |
670 | |
671 | /** |
672 | * \fn bool cbor_value_is_boolean(const CborValue *value) |
673 | * |
674 | * Returns true if the iterator \a value is valid and points to a CBOR boolean |
675 | * type (true or false). |
676 | * |
677 | * \sa cbor_value_is_valid(), cbor_value_get_boolean() |
678 | */ |
679 | |
680 | /** |
681 | * \fn CborError cbor_value_get_boolean(const CborValue *value, bool *result) |
682 | * |
683 | * Retrieves the boolean value that \a value points to and stores it in \a |
684 | * result. If the iterator \a value does not point to a boolean value, the |
685 | * behavior is undefined, so checking with \ref cbor_value_get_type or with |
686 | * \ref cbor_value_is_boolean is recommended. |
687 | * |
688 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_boolean() |
689 | */ |
690 | |
691 | /** |
692 | * \fn bool cbor_value_is_simple_type(const CborValue *value) |
693 | * |
694 | * Returns true if the iterator \a value is valid and points to a CBOR Simple Type |
695 | * type (other than true, false, null and undefined). |
696 | * |
697 | * \sa cbor_value_is_valid(), cbor_value_get_simple_type() |
698 | */ |
699 | |
700 | /** |
701 | * \fn CborError cbor_value_get_simple_type(const CborValue *value, uint8_t *result) |
702 | * |
703 | * Retrieves the CBOR Simple Type value that \a value points to and stores it |
704 | * in \a result. If the iterator \a value does not point to a simple_type |
705 | * value, the behavior is undefined, so checking with \ref cbor_value_get_type |
706 | * or with \ref cbor_value_is_simple_type is recommended. |
707 | * |
708 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_simple_type() |
709 | */ |
710 | |
711 | /** |
712 | * \fn bool cbor_value_is_integer(const CborValue *value) |
713 | * |
714 | * Returns true if the iterator \a value is valid and points to a CBOR integer |
715 | * type. |
716 | * |
717 | * \sa cbor_value_is_valid(), cbor_value_get_int, cbor_value_get_int64, cbor_value_get_uint64, cbor_value_get_raw_integer |
718 | */ |
719 | |
720 | /** |
721 | * \fn bool cbor_value_is_unsigned_integer(const CborValue *value) |
722 | * |
723 | * Returns true if the iterator \a value is valid and points to a CBOR unsigned |
724 | * integer type (positive values or zero). |
725 | * |
726 | * \sa cbor_value_is_valid(), cbor_value_get_uint64() |
727 | */ |
728 | |
729 | /** |
730 | * \fn bool cbor_value_is_negative_integer(const CborValue *value) |
731 | * |
732 | * Returns true if the iterator \a value is valid and points to a CBOR negative |
733 | * integer type. |
734 | * |
735 | * \sa cbor_value_is_valid(), cbor_value_get_int, cbor_value_get_int64, cbor_value_get_raw_integer |
736 | */ |
737 | |
738 | /** |
739 | * \fn CborError cbor_value_get_int(const CborValue *value, int *result) |
740 | * |
741 | * Retrieves the CBOR integer value that \a value points to and stores it in \a |
742 | * result. If the iterator \a value does not point to an integer value, the |
743 | * behavior is undefined, so checking with \ref cbor_value_get_type or with |
744 | * \ref cbor_value_is_integer is recommended. |
745 | * |
746 | * Note that this function does not do range-checking: integral values that do |
747 | * not fit in a variable of type \c{int} are silently truncated to fit. Use |
748 | * cbor_value_get_int_checked() if that is not acceptable. |
749 | * |
750 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_integer() |
751 | */ |
752 | |
753 | /** |
754 | * \fn CborError cbor_value_get_int64(const CborValue *value, int64_t *result) |
755 | * |
756 | * Retrieves the CBOR integer value that \a value points to and stores it in \a |
757 | * result. If the iterator \a value does not point to an integer value, the |
758 | * behavior is undefined, so checking with \ref cbor_value_get_type or with |
759 | * \ref cbor_value_is_integer is recommended. |
760 | * |
761 | * Note that this function does not do range-checking: integral values that do |
762 | * not fit in a variable of type \c{int64_t} are silently truncated to fit. Use |
763 | * cbor_value_get_int64_checked() that is not acceptable. |
764 | * |
765 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_integer() |
766 | */ |
767 | |
768 | /** |
769 | * \fn CborError cbor_value_get_uint64(const CborValue *value, uint64_t *result) |
770 | * |
771 | * Retrieves the CBOR integer value that \a value points to and stores it in \a |
772 | * result. If the iterator \a value does not point to an unsigned integer |
773 | * value, the behavior is undefined, so checking with \ref cbor_value_get_type |
774 | * or with \ref cbor_value_is_unsigned_integer is recommended. |
775 | * |
776 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_unsigned_integer() |
777 | */ |
778 | |
779 | /** |
780 | * \fn CborError cbor_value_get_raw_integer(const CborValue *value, uint64_t *result) |
781 | * |
782 | * Retrieves the CBOR integer value that \a value points to and stores it in \a |
783 | * result. If the iterator \a value does not point to an integer value, the |
784 | * behavior is undefined, so checking with \ref cbor_value_get_type or with |
785 | * \ref cbor_value_is_integer is recommended. |
786 | * |
787 | * This function is provided because CBOR negative integers can assume values |
788 | * that cannot be represented with normal 64-bit integer variables. |
789 | * |
790 | * If the integer is unsigned (that is, if cbor_value_is_unsigned_integer() |
791 | * returns true), then \a result will contain the actual value. If the integer |
792 | * is negative, then \a result will contain the absolute value of that integer, |
793 | * minus one. That is, \c {actual = -result - 1}. On architectures using two's |
794 | * complement for representation of negative integers, it is equivalent to say |
795 | * that \a result will contain the bitwise negation of the actual value. |
796 | * |
797 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_integer() |
798 | */ |
799 | |
800 | /** |
801 | * Retrieves the CBOR integer value that \a value points to and stores it in \a |
802 | * result. If the iterator \a value does not point to an integer value, the |
803 | * behavior is undefined, so checking with \ref cbor_value_get_type or with |
804 | * \ref cbor_value_is_integer is recommended. |
805 | * |
806 | * Unlike \ref cbor_value_get_int64(), this function performs a check to see if the |
807 | * stored integer fits in \a result without data loss. If the number is outside |
808 | * the valid range for the data type, this function returns the recoverable |
809 | * error CborErrorDataTooLarge. In that case, use either |
810 | * cbor_value_get_uint64() (if the number is positive) or |
811 | * cbor_value_get_raw_integer(). |
812 | * |
813 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_integer(), cbor_value_get_int64() |
814 | */ |
815 | CborError cbor_value_get_int64_checked(const CborValue *value, int64_t *result) |
816 | { |
817 | uint64_t v; |
818 | cbor_assert(cbor_value_is_integer(value)); |
819 | v = _cbor_value_extract_int64_helper(value); |
820 | |
821 | /* Check before converting, as the standard says (C11 6.3.1.3 paragraph 3): |
822 | * "[if] the new type is signed and the value cannot be represented in it; either the |
823 | * result is implementation-defined or an implementation-defined signal is raised." |
824 | * |
825 | * The range for int64_t is -2^63 to 2^63-1 (int64_t is required to be |
826 | * two's complement, C11 7.20.1.1 paragraph 3), which in CBOR is |
827 | * represented the same way, differing only on the "sign bit" (the major |
828 | * type). |
829 | */ |
830 | |
831 | if (unlikely(v > (uint64_t)INT64_MAX)) |
832 | return CborErrorDataTooLarge; |
833 | |
834 | *result = v; |
835 | if (value->flags & CborIteratorFlag_NegativeInteger) |
836 | *result = -*result - 1; |
837 | return CborNoError; |
838 | } |
839 | |
840 | /** |
841 | * Retrieves the CBOR integer value that \a value points to and stores it in \a |
842 | * result. If the iterator \a value does not point to an integer value, the |
843 | * behavior is undefined, so checking with \ref cbor_value_get_type or with |
844 | * \ref cbor_value_is_integer is recommended. |
845 | * |
846 | * Unlike \ref cbor_value_get_int(), this function performs a check to see if the |
847 | * stored integer fits in \a result without data loss. If the number is outside |
848 | * the valid range for the data type, this function returns the recoverable |
849 | * error CborErrorDataTooLarge. In that case, use one of the other integer |
850 | * functions to obtain the value. |
851 | * |
852 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_integer(), cbor_value_get_int64(), |
853 | * cbor_value_get_uint64(), cbor_value_get_int64_checked(), cbor_value_get_raw_integer() |
854 | */ |
855 | CborError cbor_value_get_int_checked(const CborValue *value, int *result) |
856 | { |
857 | uint64_t v; |
858 | cbor_assert(cbor_value_is_integer(value)); |
859 | v = _cbor_value_extract_int64_helper(value); |
860 | |
861 | /* Check before converting, as the standard says (C11 6.3.1.3 paragraph 3): |
862 | * "[if] the new type is signed and the value cannot be represented in it; either the |
863 | * result is implementation-defined or an implementation-defined signal is raised." |
864 | * |
865 | * But we can convert from signed to unsigned without fault (paragraph 2). |
866 | * |
867 | * The range for int is implementation-defined and int is not guaranteed to use |
868 | * two's complement representation (although int32_t is). |
869 | */ |
870 | |
871 | if (value->flags & CborIteratorFlag_NegativeInteger) { |
872 | if (unlikely(v > (unsigned) -(INT_MIN + 1))) |
873 | return CborErrorDataTooLarge; |
874 | |
875 | *result = (int)v; |
876 | *result = -*result - 1; |
877 | } else { |
878 | if (unlikely(v > (uint64_t)INT_MAX)) |
879 | return CborErrorDataTooLarge; |
880 | |
881 | *result = (int)v; |
882 | } |
883 | return CborNoError; |
884 | |
885 | } |
886 | |
887 | /** |
888 | * \fn bool cbor_value_is_length_known(const CborValue *value) |
889 | * |
890 | * Returns true if the length of this type is known without calculation. That |
891 | * is, if the length of this CBOR string, map or array is encoded in the data |
892 | * stream, this function returns true. If the length is not encoded, it returns |
893 | * false. |
894 | * |
895 | * If the length is known, code can call cbor_value_get_string_length(), |
896 | * cbor_value_get_array_length() or cbor_value_get_map_length() to obtain the |
897 | * length. If the length is not known but is necessary, code can use the |
898 | * cbor_value_calculate_string_length() function (no equivalent function is |
899 | * provided for maps and arrays). |
900 | */ |
901 | |
902 | /** |
903 | * \fn bool cbor_value_is_text_string(const CborValue *value) |
904 | * |
905 | * Returns true if the iterator \a value is valid and points to a CBOR text |
906 | * string. CBOR text strings are UTF-8 encoded and usually contain |
907 | * human-readable text. |
908 | * |
909 | * \sa cbor_value_is_valid(), cbor_value_get_string_length(), cbor_value_calculate_string_length(), |
910 | * cbor_value_copy_text_string(), cbor_value_dup_text_string() |
911 | */ |
912 | |
913 | /** |
914 | * \fn bool cbor_value_is_byte_string(const CborValue *value) |
915 | * |
916 | * Returns true if the iterator \a value is valid and points to a CBOR text |
917 | * string. CBOR byte strings are binary data with no specified encoding or |
918 | * format. |
919 | * |
920 | * \sa cbor_value_is_valid(), cbor_value_get_string_length(), cbor_value_calculate_string_length(), |
921 | * cbor_value_copy_byte_string(), cbor_value_dup_byte_string() |
922 | */ |
923 | |
924 | /** |
925 | * \fn CborError cbor_value_get_string_length(const CborValue *value, size_t *length) |
926 | * |
927 | * Extracts the length of the byte or text string that \a value points to and |
928 | * stores it in \a result. If the iterator \a value does not point to a text |
929 | * string or a byte string, the behaviour is undefined, so checking with \ref |
930 | * cbor_value_get_type, with \ref cbor_value_is_text_string or \ref |
931 | * cbor_value_is_byte_string is recommended. |
932 | * |
933 | * If the length of this string is not encoded in the CBOR data stream, this |
934 | * function will return the recoverable error CborErrorUnknownLength. You may |
935 | * also check whether that is the case by using cbor_value_is_length_known(). |
936 | * |
937 | * If the length of the string is required but the length was not encoded, use |
938 | * cbor_value_calculate_string_length(), but note that that function does not |
939 | * run in constant time. |
940 | * |
941 | * \note On 32-bit platforms, this function will return error condition of \ref |
942 | * CborErrorDataTooLarge if the stream indicates a length that is too big to |
943 | * fit in 32-bit. |
944 | * |
945 | * \sa cbor_value_is_valid(), cbor_value_is_length_known(), cbor_value_calculate_string_length() |
946 | */ |
947 | |
948 | /** |
949 | * Calculates the length of the byte or text string that \a value points to and |
950 | * stores it in \a len. If the iterator \a value does not point to a text |
951 | * string or a byte string, the behaviour is undefined, so checking with \ref |
952 | * cbor_value_get_type, with \ref cbor_value_is_text_string or \ref |
953 | * cbor_value_is_byte_string is recommended. |
954 | * |
955 | * This function is different from cbor_value_get_string_length() in that it |
956 | * calculates the length even for strings sent in chunks. For that reason, this |
957 | * function may not run in constant time (it will run in O(n) time on the |
958 | * number of chunks). It does use constant memory (O(1)). |
959 | * |
960 | * \note On 32-bit platforms, this function will return error condition of \ref |
961 | * CborErrorDataTooLarge if the stream indicates a length that is too big to |
962 | * fit in 32-bit. |
963 | * |
964 | * \sa cbor_value_get_string_length(), cbor_value_copy_text_string(), cbor_value_copy_byte_string(), cbor_value_is_length_known() |
965 | */ |
966 | CborError cbor_value_calculate_string_length(const CborValue *value, size_t *len) |
967 | { |
968 | *len = SIZE_MAX; |
969 | return _cbor_value_copy_string(value, NULL, buflen: len, NULL); |
970 | } |
971 | |
972 | CborError _cbor_value_begin_string_iteration(CborValue *it) |
973 | { |
974 | it->flags |= CborIteratorFlag_IteratingStringChunks | |
975 | CborIteratorFlag_BeforeFirstStringChunk; |
976 | if (!cbor_value_is_length_known(value: it)) { |
977 | /* chunked string: we're before the first chunk; |
978 | * advance to the first chunk */ |
979 | advance_bytes(it, n: 1); |
980 | } |
981 | |
982 | return CborNoError; |
983 | } |
984 | |
985 | CborError _cbor_value_finish_string_iteration(CborValue *it) |
986 | { |
987 | if (!cbor_value_is_length_known(value: it)) |
988 | advance_bytes(it, n: 1); /* skip the Break */ |
989 | |
990 | return preparse_next_value(it); |
991 | } |
992 | |
993 | static CborError get_string_chunk_size(const CborValue *it, size_t *offset, size_t *len) |
994 | { |
995 | uint8_t descriptor; |
996 | size_t bytesNeeded = 1; |
997 | |
998 | if (cbor_value_is_length_known(value: it) && (it->flags & CborIteratorFlag_BeforeFirstStringChunk) == 0) |
999 | return CborErrorNoMoreStringChunks; |
1000 | |
1001 | /* are we at the end? */ |
1002 | if (!read_bytes(it, dst: &descriptor, offset: 0, n: 1)) |
1003 | return CborErrorUnexpectedEOF; |
1004 | |
1005 | if (descriptor == BreakByte) |
1006 | return CborErrorNoMoreStringChunks; |
1007 | if ((descriptor & MajorTypeMask) != it->type) |
1008 | return CborErrorIllegalType; |
1009 | |
1010 | /* find the string length */ |
1011 | descriptor &= SmallValueMask; |
1012 | if (descriptor < Value8Bit) { |
1013 | *len = descriptor; |
1014 | } else if (unlikely(descriptor > Value64Bit)) { |
1015 | return CborErrorIllegalNumber; |
1016 | } else { |
1017 | uint64_t val; |
1018 | bytesNeeded = (size_t)(1 << (descriptor - Value8Bit)); |
1019 | if (!can_read_bytes(it, n: 1 + bytesNeeded)) |
1020 | return CborErrorUnexpectedEOF; |
1021 | |
1022 | if (descriptor <= Value16Bit) { |
1023 | if (descriptor == Value16Bit) |
1024 | val = read_uint16(it, offset: 1); |
1025 | else |
1026 | val = read_uint8(it, offset: 1); |
1027 | } else { |
1028 | if (descriptor == Value32Bit) |
1029 | val = read_uint32(it, offset: 1); |
1030 | else |
1031 | val = read_uint64(it, offset: 1); |
1032 | } |
1033 | |
1034 | *len = val; |
1035 | if (*len != val) |
1036 | return CborErrorDataTooLarge; |
1037 | |
1038 | ++bytesNeeded; |
1039 | } |
1040 | |
1041 | *offset = bytesNeeded; |
1042 | return CborNoError; |
1043 | } |
1044 | |
1045 | CborError _cbor_value_get_string_chunk_size(const CborValue *value, size_t *len) |
1046 | { |
1047 | size_t offset; |
1048 | return get_string_chunk_size(it: value, offset: &offset, len); |
1049 | } |
1050 | |
1051 | static CborError get_string_chunk(CborValue *it, const void **bufferptr, size_t *len) |
1052 | { |
1053 | size_t offset; |
1054 | CborError err = get_string_chunk_size(it, offset: &offset, len); |
1055 | if (err) |
1056 | return err; |
1057 | |
1058 | /* we're good, transfer the string now */ |
1059 | err = transfer_string(it, ptr: bufferptr, offset, len: *len); |
1060 | if (err) |
1061 | return err; |
1062 | |
1063 | /* we've iterated at least once */ |
1064 | it->flags &= ~CborIteratorFlag_BeforeFirstStringChunk; |
1065 | return CborNoError; |
1066 | } |
1067 | |
1068 | /** |
1069 | * \fn CborError cbor_value_get_text_string_chunk(const CborValue *value, const char **bufferptr, size_t *len, CborValue *next) |
1070 | * |
1071 | * Extracts one text string chunk pointed to by \a value and stores a pointer |
1072 | * to the data in \a buffer and the size in \a len, which must not be null. If |
1073 | * no more chunks are available, then \a bufferptr will be set to null. This |
1074 | * function may be used to iterate over any string without causing its contents |
1075 | * to be copied to a separate buffer, like the convenience function |
1076 | * cbor_value_copy_text_string() does. |
1077 | * |
1078 | * It is designed to be used in code like: |
1079 | * |
1080 | * \code |
1081 | * if (cbor_value_is_text_string(value)) { |
1082 | * char *ptr; |
1083 | * size_t len; |
1084 | * while (1) { |
1085 | * err = cbor_value_get_text_string_chunk(value, &ptr, &len, &value)); |
1086 | * if (err) return err; |
1087 | * if (ptr == NULL) return CborNoError; |
1088 | * consume(ptr, len); |
1089 | * } |
1090 | * } |
1091 | * \endcode |
1092 | * |
1093 | * If the iterator \a value does not point to a text string, the behaviour is |
1094 | * undefined, so checking with \ref cbor_value_get_type or \ref |
1095 | * cbor_value_is_text_string is recommended. |
1096 | * |
1097 | * The \a next pointer, if not null, will be updated to point to the next item |
1098 | * after this string. During iteration, the pointer must only be passed back |
1099 | * again to this function; passing it to any other function in this library |
1100 | * results in undefined behavior. If there are no more chunks to be read from |
1101 | * \a value, then \a next will be set to the next item after this string; if \a |
1102 | * value points to the last item, then \a next will be invalid. |
1103 | * |
1104 | * \note This function does not perform UTF-8 validation on the incoming text |
1105 | * string. |
1106 | * |
1107 | * \sa cbor_value_dup_text_string(), cbor_value_copy_text_string(), cbor_value_caculate_string_length(), cbor_value_get_byte_string_chunk() |
1108 | */ |
1109 | |
1110 | /** |
1111 | * \fn CborError cbor_value_get_byte_string_chunk(const CborValue *value, const char **bufferptr, size_t *len, CborValue *next) |
1112 | * |
1113 | * Extracts one byte string chunk pointed to by \a value and stores a pointer |
1114 | * to the data in \a buffer and the size in \a len, which must not be null. If |
1115 | * no more chunks are available, then \a bufferptr will be set to null. This |
1116 | * function may be used to iterate over any string without causing its contents |
1117 | * to be copied to a separate buffer, like the convenience function |
1118 | * cbor_value_copy_byte_string() does. |
1119 | * |
1120 | * It is designed to be used in code like: |
1121 | * |
1122 | * \code |
1123 | * if (cbor_value_is_byte_string(value)) { |
1124 | * char *ptr; |
1125 | * size_t len; |
1126 | * while (1) { |
1127 | * err = cbor_value_get_byte_string_chunk(value, &ptr, &len, &value)); |
1128 | * if (err) return err; |
1129 | * if (ptr == NULL) return CborNoError; |
1130 | * consume(ptr, len); |
1131 | * } |
1132 | * } |
1133 | * \endcode |
1134 | * |
1135 | * If the iterator \a value does not point to a byte string, the behaviour is |
1136 | * undefined, so checking with \ref cbor_value_get_type or \ref |
1137 | * cbor_value_is_byte_string is recommended. |
1138 | * |
1139 | * The \a next pointer, if not null, will be updated to point to the next item |
1140 | * after this string. During iteration, the pointer must only be passed back |
1141 | * again to this function; passing it to any other function in this library |
1142 | * results in undefined behavior. If there are no more chunks to be read from |
1143 | * \a value, then \a next will be set to the next item after this string; if \a |
1144 | * value points to the last item, then \a next will be invalid. |
1145 | * |
1146 | * \sa cbor_value_dup_byte_string(), cbor_value_copy_byte_string(), cbor_value_caculate_string_length(), cbor_value_get_text_string_chunk() |
1147 | */ |
1148 | |
1149 | CborError _cbor_value_get_string_chunk(const CborValue *value, const void **bufferptr, |
1150 | size_t *len, CborValue *next) |
1151 | { |
1152 | CborValue tmp; |
1153 | if (!next) |
1154 | next = &tmp; |
1155 | *next = *value; |
1156 | return get_string_chunk(it: next, bufferptr, len); |
1157 | } |
1158 | |
1159 | /* We return uintptr_t so that we can pass memcpy directly as the iteration |
1160 | * function. The choice is to optimize for memcpy, which is used in the base |
1161 | * parser API (cbor_value_copy_string), while memcmp is used in convenience API |
1162 | * only. */ |
1163 | typedef uintptr_t (*IterateFunction)(char *, const uint8_t *, size_t); |
1164 | |
1165 | static uintptr_t iterate_noop(char *dest, const uint8_t *src, size_t len) |
1166 | { |
1167 | (void)dest; |
1168 | (void)src; |
1169 | (void)len; |
1170 | return true; |
1171 | } |
1172 | |
1173 | static uintptr_t iterate_memcmp(char *s1, const uint8_t *s2, size_t len) |
1174 | { |
1175 | return memcmp(s1: s1, s2: (const char *)s2, n: len) == 0; |
1176 | } |
1177 | |
1178 | static uintptr_t iterate_memcpy(char *dest, const uint8_t *src, size_t len) |
1179 | { |
1180 | return (uintptr_t)memcpy(dest: dest, src: src, n: len); |
1181 | } |
1182 | |
1183 | static CborError iterate_string_chunks(const CborValue *value, char *buffer, size_t *buflen, |
1184 | bool *result, CborValue *next, IterateFunction func) |
1185 | { |
1186 | CborError err; |
1187 | CborValue tmp; |
1188 | size_t total = 0; |
1189 | const void *ptr; |
1190 | |
1191 | cbor_assert(cbor_value_is_byte_string(value) || cbor_value_is_text_string(value)); |
1192 | if (!next) |
1193 | next = &tmp; |
1194 | *next = *value; |
1195 | *result = true; |
1196 | |
1197 | err = _cbor_value_begin_string_iteration(it: next); |
1198 | if (err) |
1199 | return err; |
1200 | |
1201 | while (1) { |
1202 | size_t newTotal; |
1203 | size_t chunkLen; |
1204 | err = get_string_chunk(it: next, bufferptr: &ptr, len: &chunkLen); |
1205 | if (err == CborErrorNoMoreStringChunks) |
1206 | break; |
1207 | if (err) |
1208 | return err; |
1209 | |
1210 | if (unlikely(add_check_overflow(total, chunkLen, &newTotal))) |
1211 | return CborErrorDataTooLarge; |
1212 | |
1213 | if (*result && *buflen >= newTotal) |
1214 | *result = !!func(buffer + total, (const uint8_t *)ptr, chunkLen); |
1215 | else |
1216 | *result = false; |
1217 | |
1218 | total = newTotal; |
1219 | } |
1220 | |
1221 | /* is there enough room for the ending NUL byte? */ |
1222 | if (*result && *buflen > total) { |
1223 | uint8_t nul[] = { 0 }; |
1224 | *result = !!func(buffer + total, nul, 1); |
1225 | } |
1226 | *buflen = total; |
1227 | return _cbor_value_finish_string_iteration(it: next); |
1228 | } |
1229 | |
1230 | /** |
1231 | * \fn CborError cbor_value_copy_text_string(const CborValue *value, char *buffer, size_t *buflen, CborValue *next) |
1232 | * |
1233 | * Copies the string pointed to by \a value into the buffer provided at \a buffer |
1234 | * of \a buflen bytes. If \a buffer is a NULL pointer, this function will not |
1235 | * copy anything and will only update the \a next value. |
1236 | * |
1237 | * If the iterator \a value does not point to a text string, the behaviour is |
1238 | * undefined, so checking with \ref cbor_value_get_type or \ref |
1239 | * cbor_value_is_text_string is recommended. |
1240 | * |
1241 | * If the provided buffer length was too small, this function returns an error |
1242 | * condition of \ref CborErrorOutOfMemory. If you need to calculate the length |
1243 | * of the string in order to preallocate a buffer, use |
1244 | * cbor_value_calculate_string_length(). |
1245 | * |
1246 | * On success, this function sets the number of bytes copied to \c{*buflen}. If |
1247 | * the buffer is large enough, this function will insert a null byte after the |
1248 | * last copied byte, to facilitate manipulation of text strings. That byte is |
1249 | * not included in the returned value of \c{*buflen}. If there was no space for |
1250 | * the terminating null, no error is returned, so callers must check the value |
1251 | * of *buflen after the call, before relying on the '\0'; if it has not been |
1252 | * changed by the call, there is no '\0'-termination on the buffer's contents. |
1253 | * |
1254 | * The \a next pointer, if not null, will be updated to point to the next item |
1255 | * after this string. If \a value points to the last item, then \a next will be |
1256 | * invalid. |
1257 | * |
1258 | * This function may not run in constant time (it will run in O(n) time on the |
1259 | * number of chunks). It requires constant memory (O(1)). |
1260 | * |
1261 | * \note This function does not perform UTF-8 validation on the incoming text |
1262 | * string. |
1263 | * |
1264 | * \sa cbor_value_get_text_string_chunk() cbor_value_dup_text_string(), cbor_value_copy_byte_string(), cbor_value_get_string_length(), cbor_value_calculate_string_length() |
1265 | */ |
1266 | |
1267 | /** |
1268 | * \fn CborError cbor_value_copy_byte_string(const CborValue *value, uint8_t *buffer, size_t *buflen, CborValue *next) |
1269 | * |
1270 | * Copies the string pointed by \a value into the buffer provided at \a buffer |
1271 | * of \a buflen bytes. If \a buffer is a NULL pointer, this function will not |
1272 | * copy anything and will only update the \a next value. |
1273 | * |
1274 | * If the iterator \a value does not point to a byte string, the behaviour is |
1275 | * undefined, so checking with \ref cbor_value_get_type or \ref |
1276 | * cbor_value_is_byte_string is recommended. |
1277 | * |
1278 | * If the provided buffer length was too small, this function returns an error |
1279 | * condition of \ref CborErrorOutOfMemory. If you need to calculate the length |
1280 | * of the string in order to preallocate a buffer, use |
1281 | * cbor_value_calculate_string_length(). |
1282 | * |
1283 | * On success, this function sets the number of bytes copied to \c{*buflen}. If |
1284 | * the buffer is large enough, this function will insert a null byte after the |
1285 | * last copied byte, to facilitate manipulation of null-terminated strings. |
1286 | * That byte is not included in the returned value of \c{*buflen}. |
1287 | * |
1288 | * The \a next pointer, if not null, will be updated to point to the next item |
1289 | * after this string. If \a value points to the last item, then \a next will be |
1290 | * invalid. |
1291 | * |
1292 | * This function may not run in constant time (it will run in O(n) time on the |
1293 | * number of chunks). It requires constant memory (O(1)). |
1294 | * |
1295 | * \sa cbor_value_get_byte_string_chunk(), cbor_value_dup_text_string(), cbor_value_copy_text_string(), cbor_value_get_string_length(), cbor_value_calculate_string_length() |
1296 | */ |
1297 | |
1298 | CborError _cbor_value_copy_string(const CborValue *value, void *buffer, |
1299 | size_t *buflen, CborValue *next) |
1300 | { |
1301 | bool copied_all; |
1302 | CborError err = iterate_string_chunks(value, buffer: (char*)buffer, buflen, result: &copied_all, next, |
1303 | func: buffer ? iterate_memcpy : iterate_noop); |
1304 | return err ? err : |
1305 | copied_all ? CborNoError : CborErrorOutOfMemory; |
1306 | } |
1307 | |
1308 | /** |
1309 | * Compares the entry \a value with the string \a string and stores the result |
1310 | * in \a result. If the value is different from \a string \a result will |
1311 | * contain \c false. |
1312 | * |
1313 | * The entry at \a value may be a tagged string. If \a value is not a string or |
1314 | * a tagged string, the comparison result will be false. |
1315 | * |
1316 | * CBOR requires text strings to be encoded in UTF-8, but this function does |
1317 | * not validate either the strings in the stream or the string \a string to be |
1318 | * matched. Moreover, comparison is done on strict codepoint comparison, |
1319 | * without any Unicode normalization. |
1320 | * |
1321 | * This function may not run in constant time (it will run in O(n) time on the |
1322 | * number of chunks). It requires constant memory (O(1)). |
1323 | * |
1324 | * \sa cbor_value_skip_tag(), cbor_value_copy_text_string() |
1325 | */ |
1326 | CborError cbor_value_text_string_equals(const CborValue *value, const char *string, bool *result) |
1327 | { |
1328 | size_t len; |
1329 | CborValue copy = *value; |
1330 | CborError err = cbor_value_skip_tag(it: ©); |
1331 | if (err) |
1332 | return err; |
1333 | if (!cbor_value_is_text_string(value: ©)) { |
1334 | *result = false; |
1335 | return CborNoError; |
1336 | } |
1337 | |
1338 | len = strlen(s: string); |
1339 | return iterate_string_chunks(value: ©, CONST_CAST(char *, string), buflen: &len, result, NULL, func: iterate_memcmp); |
1340 | } |
1341 | |
1342 | /** |
1343 | * \fn bool cbor_value_is_array(const CborValue *value) |
1344 | * |
1345 | * Returns true if the iterator \a value is valid and points to a CBOR array. |
1346 | * |
1347 | * \sa cbor_value_is_valid(), cbor_value_is_map() |
1348 | */ |
1349 | |
1350 | /** |
1351 | * \fn CborError cbor_value_get_array_length(const CborValue *value, size_t *length) |
1352 | * |
1353 | * Extracts the length of the CBOR array that \a value points to and stores it |
1354 | * in \a result. If the iterator \a value does not point to a CBOR array, the |
1355 | * behaviour is undefined, so checking with \ref cbor_value_get_type or \ref |
1356 | * cbor_value_is_array is recommended. |
1357 | * |
1358 | * If the length of this array is not encoded in the CBOR data stream, this |
1359 | * function will return the recoverable error CborErrorUnknownLength. You may |
1360 | * also check whether that is the case by using cbor_value_is_length_known(). |
1361 | * |
1362 | * \note On 32-bit platforms, this function will return error condition of \ref |
1363 | * CborErrorDataTooLarge if the stream indicates a length that is too big to |
1364 | * fit in 32-bit. |
1365 | * |
1366 | * \sa cbor_value_is_valid(), cbor_value_is_length_known() |
1367 | */ |
1368 | |
1369 | /** |
1370 | * \fn bool cbor_value_is_map(const CborValue *value) |
1371 | * |
1372 | * Returns true if the iterator \a value is valid and points to a CBOR map. |
1373 | * |
1374 | * \sa cbor_value_is_valid(), cbor_value_is_array() |
1375 | */ |
1376 | |
1377 | /** |
1378 | * \fn CborError cbor_value_get_map_length(const CborValue *value, size_t *length) |
1379 | * |
1380 | * Extracts the length of the CBOR map that \a value points to and stores it in |
1381 | * \a result. If the iterator \a value does not point to a CBOR map, the |
1382 | * behaviour is undefined, so checking with \ref cbor_value_get_type or \ref |
1383 | * cbor_value_is_map is recommended. |
1384 | * |
1385 | * If the length of this map is not encoded in the CBOR data stream, this |
1386 | * function will return the recoverable error CborErrorUnknownLength. You may |
1387 | * also check whether that is the case by using cbor_value_is_length_known(). |
1388 | * |
1389 | * \note On 32-bit platforms, this function will return error condition of \ref |
1390 | * CborErrorDataTooLarge if the stream indicates a length that is too big to |
1391 | * fit in 32-bit. |
1392 | * |
1393 | * \sa cbor_value_is_valid(), cbor_value_is_length_known() |
1394 | */ |
1395 | |
1396 | /** |
1397 | * Attempts to find the value in map \a map that corresponds to the text string |
1398 | * entry \a string. If the iterator \a value does not point to a CBOR map, the |
1399 | * behaviour is undefined, so checking with \ref cbor_value_get_type or \ref |
1400 | * cbor_value_is_map is recommended. |
1401 | * |
1402 | * If the item is found, it is stored in \a result. If no item is found |
1403 | * matching the key, then \a result will contain an element of type \ref |
1404 | * CborInvalidType. Matching is performed using |
1405 | * cbor_value_text_string_equals(), so tagged strings will also match. |
1406 | * |
1407 | * This function has a time complexity of O(n) where n is the number of |
1408 | * elements in the map to be searched. In addition, this function is has O(n) |
1409 | * memory requirement based on the number of nested containers (maps or arrays) |
1410 | * found as elements of this map. |
1411 | * |
1412 | * \sa cbor_value_is_valid(), cbor_value_text_string_equals(), cbor_value_advance() |
1413 | */ |
1414 | CborError cbor_value_map_find_value(const CborValue *map, const char *string, CborValue *element) |
1415 | { |
1416 | CborError err; |
1417 | size_t len = strlen(s: string); |
1418 | cbor_assert(cbor_value_is_map(map)); |
1419 | err = cbor_value_enter_container(it: map, recursed: element); |
1420 | if (err) |
1421 | goto error; |
1422 | |
1423 | while (!cbor_value_at_end(it: element)) { |
1424 | /* find the non-tag so we can compare */ |
1425 | err = cbor_value_skip_tag(it: element); |
1426 | if (err) |
1427 | goto error; |
1428 | if (cbor_value_is_text_string(value: element)) { |
1429 | bool equals; |
1430 | size_t dummyLen = len; |
1431 | err = iterate_string_chunks(value: element, CONST_CAST(char *, string), buflen: &dummyLen, |
1432 | result: &equals, next: element, func: iterate_memcmp); |
1433 | if (err) |
1434 | goto error; |
1435 | if (equals) |
1436 | return preparse_value(it: element); |
1437 | } else { |
1438 | /* skip this key */ |
1439 | err = cbor_value_advance(it: element); |
1440 | if (err) |
1441 | goto error; |
1442 | } |
1443 | |
1444 | /* skip this value */ |
1445 | err = cbor_value_skip_tag(it: element); |
1446 | if (err) |
1447 | goto error; |
1448 | err = cbor_value_advance(it: element); |
1449 | if (err) |
1450 | goto error; |
1451 | } |
1452 | |
1453 | /* not found */ |
1454 | element->type = CborInvalidType; |
1455 | return CborNoError; |
1456 | |
1457 | error: |
1458 | element->type = CborInvalidType; |
1459 | return err; |
1460 | } |
1461 | |
1462 | /** |
1463 | * \fn bool cbor_value_is_float(const CborValue *value) |
1464 | * |
1465 | * Returns true if the iterator \a value is valid and points to a CBOR |
1466 | * single-precision floating point (32-bit). |
1467 | * |
1468 | * \sa cbor_value_is_valid(), cbor_value_is_double(), cbor_value_is_half_float() |
1469 | */ |
1470 | |
1471 | /** |
1472 | * \fn CborError cbor_value_get_float(const CborValue *value, float *result) |
1473 | * |
1474 | * Retrieves the CBOR single-precision floating point (32-bit) value that \a |
1475 | * value points to and stores it in \a result. If the iterator \a value does |
1476 | * not point to a single-precision floating point value, the behavior is |
1477 | * undefined, so checking with \ref cbor_value_get_type or with \ref |
1478 | * cbor_value_is_float is recommended. |
1479 | * |
1480 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_float(), cbor_value_get_double() |
1481 | */ |
1482 | |
1483 | /** |
1484 | * \fn bool cbor_value_is_double(const CborValue *value) |
1485 | * |
1486 | * Returns true if the iterator \a value is valid and points to a CBOR |
1487 | * double-precision floating point (64-bit). |
1488 | * |
1489 | * \sa cbor_value_is_valid(), cbor_value_is_float(), cbor_value_is_half_float() |
1490 | */ |
1491 | |
1492 | /** |
1493 | * \fn CborError cbor_value_get_double(const CborValue *value, float *result) |
1494 | * |
1495 | * Retrieves the CBOR double-precision floating point (64-bit) value that \a |
1496 | * value points to and stores it in \a result. If the iterator \a value does |
1497 | * not point to a double-precision floating point value, the behavior is |
1498 | * undefined, so checking with \ref cbor_value_get_type or with \ref |
1499 | * cbor_value_is_double is recommended. |
1500 | * |
1501 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_double(), cbor_value_get_float() |
1502 | */ |
1503 | |
1504 | /** |
1505 | * \fn bool cbor_value_is_half_float(const CborValue *value) |
1506 | * |
1507 | * Returns true if the iterator \a value is valid and points to a CBOR |
1508 | * single-precision floating point (16-bit). |
1509 | * |
1510 | * \sa cbor_value_is_valid(), cbor_value_is_double(), cbor_value_is_float() |
1511 | */ |
1512 | |
1513 | /** |
1514 | * \fn CborError cbor_value_get_half_float(const CborValue *value, void *result) |
1515 | * |
1516 | * Retrieves the CBOR half-precision floating point (16-bit) value that \a |
1517 | * value points to and stores it in \a result. If the iterator \a value does |
1518 | * not point to a half-precision floating point value, the behavior is |
1519 | * undefined, so checking with \ref cbor_value_get_type or with \ref |
1520 | * cbor_value_is_half_float is recommended. |
1521 | * |
1522 | * Note: since the C language does not have a standard type for half-precision |
1523 | * floating point, this function takes a \c{void *} as a parameter for the |
1524 | * storage area, which must be at least 16 bits wide. |
1525 | * |
1526 | * \sa cbor_value_get_type(), cbor_value_is_valid(), cbor_value_is_half_float(), cbor_value_get_half_float_as_float(), cbor_value_get_float() |
1527 | */ |
1528 | |
1529 | /** @} */ |
1530 | |